University of Arizona Generative AI
(U of A GenAI)
Advanced responsible AI for research, education, and operations
Chat with state-of-the-art AI large language models (LLMs) in a secure environment designed for the U of A campus community.
How to access
Get access to the tool, U of A GenAI.
Note: During the early access phase, some features, such as guardrails, might not be enabled or may change without notice.
About U of A GenAI
The U of A GenAI tool is a modern chat interface powered by leading large language models (LLMs), designed to support research, education, operations, and general use across the University of Arizona. Freely accessible to the U of A community, the platform allows users to interact with, compare, and evaluate multiple LLMs side-by-side within a single, intuitive chat experience.
Whether you are exploring ideas, drafting content, analyzing documents, or prototyping workflows, the U of A GenAI tool gives you flexible access to advanced AI without compromising privacy or data security at no cost.
The U of A GenAI tool has not been approved for the use of 'Restricted Data' as defined by the University of Arizona's Information Classification Standard. Restricted and regulated data types such as Health Information, Credit Card Numbers, Social Security Numbers, etc., should not be entered into the GenAI tool. For questions, refer to University Privacy.
Privacy & Security
- Private conversations — Your chats are never used for training AI models, and the U of A will not access them
- Secure document upload — No public disclosure of proprietary data
Powerful Capabilities
- Multiple foundation models — Access a variety of state-of-the-art LLMs
- Model comparison — Compare outputs side-by-side within the chat UI
- Web search integration — Get up-to-date information in your conversations
Built for Campus
- Free for U of A community — No cost for students, faculty, and staff
- Easy collaboration — Share conversations with other U of A GenAI platform users
- Ready to use — Start chatting immediately with enterprise-grade AI
Quick Start Guide
Upon accessing the U of A GenAI tool, click the login button "Continue with University of Arizona"
You will be redirected to the familiar University of Arizona Web Auth system. Enter your credentials and click on "Login". You might need to approve the login with the Duo Push application.
Welcome to the University of Arizona GenAI platform! On the left-hand side, you will see an expandable menu that lets you start a new chat, access previous chats, complete searches, create folders, and add notes.
At the top of the page, you can select and expand a dropdown menu to choose the model you want to use. For example, here we are using Claude Haiku 4.5 from Anthropic. Next to the drop-down is a plus symbol that lets you add more models and chat with them all simultaneously, allowing you to compare their outputs.
After selecting a model, start chatting! You can upload files for the models to read and use, and let the model perform a web search.
To help with organization, you can create folders and move chats to them; These folders can be named and have a set System Prompt. A system prompt is a foundational instruction set given to a large language model (LLM) that defines its role, behavior, tone, constraints, and capabilities for a specific use case or application. The system prompt will be applied to all of the chats in the folder.
Once your folder is created, it will be visible in the side panel. You can drag existing conversations into the folder and start new ones.
Available LLMs & Common Uses
Works with: Text and images
Best for: Most students needs and general campus use across departments
Claude Haiku is one of the best AI models available through AWS Bedrock. It provides intelligent, quick responses perfect for academic and administrative tasks. It can handle long documents and conversations (up to 200,000 words) and offers strong analytical capabilities.
Estimated Resource Needs: $ $ $ $ $
(This is only in comparison with other LLMs listed here)
Ideal campus applications:
- Homework and assignment assistance (check with your instructor about course AI policies)
- Essay writing and research support
- Study assistance and tutoring
- Course material development
- Student advising and support services
- Administrative automation
- Image analysis (charts, diagrams, handwritten work)
Works with: Text and images
Best for: Deep reasoning and coding tasks
Estimated Resource Needs: $ $ $ $ $
(This is only in comparison with other LLMs listed here)
Claude Opus 4.6 is Anthropic’s most capable model, designed for complex, long-horizon tasks in coding, enterprise workflows, and professional knowledge work. It can handle long documents and conversations (up to 1 million words) and offers strong analytical capabilities.
Ideal campus applications:
- Complex coding tasks
- Advanced research support
- Data-driven analysis
- Course material development
- Image analysis (charts, diagrams, handwritten work)
Works with: Text and images
Best for: Multilingual applications and extended conversations
Estimated Resource Needs: $ $ $ $ $
(This is only in comparison with other LLMs listed here)
Gemma excels at maintaining long conversations (128,000 words), works with multiple languages, and handles both text and images efficiently while delivering solid performance.
Ideal campus applications:
- International student services
- Extended tutoring sessions
- Multi-language document processing
- High-volume projects
- Departmental chatbots
- Multilingual research support
Works with: Text only (no images)
Best for: Complex analytical tasks requiring advanced reasoning
Estimated Resource Needs: $ $ $ $ $
(This is only in comparison with other LLMs listed here)
This model delivers GPT-4-level performance with greater flexibility for customization. It handles long conversations well (128,000 words) and excels at reasoning and using tools, though it cannot process images.
Ideal campus applications:
- Complex analytical projects
- Advanced reasoning tasks
- Tool integration workflows
- Custom AI applications
- Research requiring GPT-4 level capability
- Text-only document analysis
Works with: Text and images
Best for: Analyzing very long documents or comprehensive research materials
Estimated Resource Needs: $ $ $ $ $
(This is only in comparison with other LLMs listed here)
Llama's standout feature is its ability to handle extremely long documents (up to 1 million words—roughly 3-4 novels). It supports multiple languages and is particularly strong at coding and technical reasoning.
Ideal campus applications:
- Dissertation and thesis analysis
- Large-scale literature reviews
- Processing entire course readers or textbooks
- Comprehensive document summarization
- Multi-chapter book analysis
- Long-form research synthesis
Works with: Text and images
Best for: Teams prioritizing AWS integration and enterprise support
Estimated Resource Needs: $ $ $ $ $
(This is only in comparison with other LLMs listed here)
As Amazon's own model, Nova Pro offers excellent integration with other AWS services used on campus. It balances good performance and handles very long contexts (300,000 words). Supports many languages.
Ideal campus applications:
- Question answering systems
- Content generation and summarization
- Knowledge base integration
- Projects already using AWS infrastructure
- RAG (Retrieval-Augmented Generation) applications
- Multi-language content workflows
Frequently Asked Questions
Yes, the University of Arizona will not access your chats or the files you upload. Your data will not be used to train any models.
No, U of A GenAI is not HIPAA compliant. It is only FERPA compliant. You may upload academic records, but not any personal health information.
Have an active University of Arizona NetID. Currently U of A GenAI is available only in limited early access.
Uploaded document files will be cached only after a certain period of time. A tool is currently in development that will allow for longer storage for more involved projects for courses and research teams.
Availability of LLMs will be changed over time, rotating older models for newer ones as they become available.
The U of A GenAI tool is built using open-source technologies, including Open WebUI and LiteLLM.
Join the Responsible AI Community
The Arizona Institute for AI and Society (AI2S) is continuously developing new secure AI-related tools to complement the U of A GenAI tool.
Watch AI tools used in helpful use cases. Join AI Insights interactive sessions that feature faculty researchers, student speakers, and industry leaders who share trends, practical applications, diverse perspectives, and emerging tools in AI.
With AI Info sessions, learn to build confidence, fluency, and shared understanding around responsible AI. Each virtual session is intended to spark curiosity, create a common language, and bring clarity to responsible AI and what “AI at work” looks like.
To receive the latest release updates, research highlights, and campus responsible AI news from both AI2S and the Office of Responsible AI, sign up to join the mailing list and our community.